Tool Use on ToolBench standard evaluation
51.3Pass RateToolRLA
Evaluation results
| Method | Links | |
|---|---|---|
| ToolRLABackbone=Qwen3-14B2026.03 | 51.3 | |
| Bloomberg AI Engineering2026.03 | 48.2 | |
| GPT-4 (function calling)2026.03 | 46.2 | |
| AvaTaR2026.03 | 44.3 | |
| ToolLLM2026.03 | 36.8 | |
| Gorilla2026.03 | 20.4 |